BIND 10 #1672: b10-resolver crashing with status 11 when compiled with --enable-boost-threads (was: b10-resolver crashing with status 11)

[BIND 10 Development]

#1672: b10-resolver crashing with status 11 when compiled with --enable-boost-
threads
-------------------------------------+-------------------------------------
                   Reporter:         |                 Owner:
  Jeffry A. Spain                    |                Status:  new
                       Type:         |             Milestone:  New Tasks
  defect                             |            Resolution:
                   Priority:  minor  |             Sensitive:  0
                  Component:         |           Sub-Project:  DNS
  resolver                           |  Estimated Difficulty:  0
                   Keywords:         |           Total Hours:  0
            Defect Severity:  Low    |
Feature Depending on Ticket:         |
        Add Hours to Ticket:  0      |
                  Internal?:  0      |
-------------------------------------+-------------------------------------
Changes (by vorner):

 * priority:  major => minor
 * severity:  High => Low


Comment:

 Thank you for your info, with the flag I can reproduce the problem (even
 the tests crash now). Core dumps are of little use without the
 corresponding binaries, but we can now get our own, so this isn't a
 problem.

 Anyway, to the flag itself, it is kind of historical relict now, I'd say
 (I wasn't aware of the flag until you poited it out). We wrote some code
 and put locks there to be ready when we start using threads and it turned
 out the boost library didn't provide the locks on all the systems. So we
 put an ifdef there and provided some dummy locks which are used instead if
 the flag isn't used. We don't use any threads yet (we just have the locks
 in place), so using the flag would only slow things down for now. It seems
 as nobody was using it, the code there got rot somehow.

 I'd suggest you use the code without the flag, then. I'll propose to
 remove the flag from configure (it brings no advantage and only breaks
 things). We'll need to solve the problem before we actually introduce the
 threads, but that probably won't happen sometime really soon.

-- 
Ticket URL: <http://bind10.isc.org/ticket/1672#comment:5>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development